PCI device support in Open Firmware (device tree syntax)

PCI device support in Open Firmware (device tree syntax)

[Johnny Hung]

Hi all,
    I am working in customized Freescale MPC8313 board. There are two
PCI devices (Broadcom Switch) in PCI bus.
Each PCI device has its configuration space. It contains
vendor/product ID (RO)..., and important information, likes BARs
(Base Address), INT line and IRQ(RW). These resources (BAR, INT, IRQ)
is assigned from BIOS in x86 arch.
    My problem is how to assign PCI device resources in device tree. I
can't find exist dts file as an example. I have read the
http://www.power.org/resources/downloads/Power_ePAPR_APPROVED_v1.0.pdf
file. It's described detailed about device tree
except PCI device. So, anyone give me a hint or sany tuff is appreciated.
   I think Linux Kernel parses device tree and get the PCI device
resources information then write into PCI configuration space.
So PCI device driver do pci_register_driver (get resource) and do
itself probe for PCI device. Is it right? CMIIAW

BRs, H. Johnny

Re: PCI device support in Open Firmware (device tree syntax)

[Scott Wood]

On Mon, Jun 29, 2009 at 11:58:22PM +0800, Johnny Hung wrote:
> Hi all,
>     I am working in customized Freescale MPC8313 board. There are two
> PCI devices (Broadcom Switch) in PCI bus.
> Each PCI device has its configuration space. It contains
> vendor/product ID (RO)..., and important information, likes BARs
> (Base Address), INT line and IRQ(RW). These resources (BAR, INT, IRQ)
> is assigned from BIOS in x86 arch.
>     My problem is how to assign PCI device resources in device tree. I
> can't find exist dts file as an example. I have read the
> http://www.power.org/resources/downloads/Power_ePAPR_APPROVED_v1.0.pdf
> file. It's described detailed about device tree
> except PCI device. So, anyone give me a hint or sany tuff is appreciated.
>    I think Linux Kernel parses device tree and get the PCI device
> resources information then write into PCI configuration space.
> So PCI device driver do pci_register_driver (get resource) and do
> itself probe for PCI device. Is it right? CMIIAW

With flat device trees, PCI devices are not typically included as they
can be probed instead.  Interrupt mapping is conveyed by the
interrupt-map property in the PCI controller node.

-Scott

Re: PCI device support in Open Firmware (device tree syntax)

[Johnny Hung]

Thanks for your reply. So there are no PCI device resource info in
flat device tree.
How do PCI device resources be assign in general case or is there any examp=
le?
I am so confusing, would you please give me detailed describle?

BRs, H. Johnny
>
> With flat device trees, PCI devices are not typically included as they
> can be probed instead. =A0Interrupt mapping is conveyed by the
> interrupt-map property in the PCI controller node.
>
> -Scott
>

Re: PCI device support in Open Firmware (device tree syntax)

[Johnny Hung]

2009/6/30 Marco Stornelli <marco.stornelli@gmail.com>:
> 2009/6/30 Johnny Hung <johnny.hacking@gmail.com>:
>> Thanks for your reply. So there are no PCI device resource info in
>> flat device tree.
>> How do PCI device resources be assign in general case or is there any example?
>> I am so confusing, would you please give me detailed describle?
>>
>> BRs, H. Johnny
>>>
>
> You should look at your bootloader about the initial configuration of
> the pci device, in general it should (more or less) do the work of
> BIOS as you said. After that the kernel can probe them.
>
> Marco
>
You are right. u-boot do PCI device BAR resources assign and flat
device tree describe PCI device interrupt-map. I use "lspci -s xxx -x"
for the PCI device, the resources was assigned properly.
Thanks.

BRs, H. Johnny

Re: PCI device support in Open Firmware (device tree syntax)

[Benjamin Herrenschmidt]

> You are right. u-boot do PCI device BAR resources assign and flat
> device tree describe PCI device interrupt-map. I use "lspci -s xxx -x"
> for the PCI device, the resources was assigned properly.

Note that the device-tree -can- contain PCI devices, but it's not
mandatory.

If your bootloader does the probing and configuration of the PCI bus, it
might be a good idea for that bootloader to also create the PCI device
nodes in the tree, since this will allow in a near future the kernel to
"instanciate" them from that tree instead of probing with config space,
which is faster and more reliably. (Currently only ppc64 does that but
it will become generic soon).

Cheers,
Ben.